Study design

The present study will conduct a series of N-of-1 superiority trials on Baduanjin exercise for the management of CNLBP. The flow diagram is shown in figure 1. First, patients will be assessed for eligibility and participate in a Baduanjin exercise training. Then, eligible participants will receive 1–3 treatment cycles. The number of experienced treatment cycles in each patient depends on the results of statistical analysis at the end of each cycle. Each cycle includes a period of Baduanjin exercise and a period of waiting list. Typically, a 1-week washout period, during which therapies for relieving CNLBP are not allowed to eliminate the efficacy of previously received interventions, will be set between the above-mentioned two periods in view of the feasibility of N-of-1 trials. It follows the Standard Protocol Items: Recommendations for Interventional Trials statement.38 Outpatients or inpatients will be recruited from the Department of Orthopedics in the First Teaching Hospital of Tianjin University of TCM. This study will be conducted from 1 October 2023 to 31 December 2025. To catch the attention of potentially eligible patients and obtain sufficient participant recruitment to reach the target sample size, recruitment advertisements will be posted at the entrance of outpatient and inpatient departments. Items of the WHO Trial Registration Data Set, registration date and protocol version are available at https://www.chictr.org.cn/showprojEN.html?proj=172369.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow diagram of a series of N-of-1 trials.

Patients

Random assignment and allocation concealment

The eligible patients will be randomised into Baduanjin exercise or waiting list group during the two periods of each treatment cycle. For example, a patient may take Baduanjin exercise during the first period but not take the exercise during the second period in a treatment cycle. The random sequence may be different across the treatment cycles. For instance, a patient may take Baduanjin exercise during the first period of the first treatment cycle but not take it during the first period of the second treatment cycle. Before these N-of-1 trials are conducted, the random allocation sequence will be generated with SAS V.9.1 software by a statistician who is not directly involved in these N-of-1 trials. The doctor (WY) will acquire the random sequence of each patient by contacting the above-mentioned statistician to manage the assignment of interventions, and inform the patients to perform the Baduanjin exercise. It means that random allocation will be concealed to WY and each patient before the initiation of each treatment cycle.

Interventions

Outcomes

In this study, the pain intensity determined using the 10-point VAS will be our primary outcome, with a higher VAS score indicating the more severe pain. The 10-point VAS displays good reliability (r=0.96) and validity (r=0.97).44 The secondary outcomes include the Oswestry Disability Index (ODI), the Japanese Orthopaedic Association Back Pain Evaluation Questionnaire (JOABPEQ) and the Short Form Health Survey 12 (SF-12). ODI is a patient-reported outcome tool, which is highly reliable (r=0.89) and valid (r=0.76).45 The value ranges from 0% to 100%,46 with the higher ODI indicating the worse physical functioning. JOABPEQ, the self-administered questionnaire, shows high reliability (r=0.977) and good validity (r=0.726).47 It consists of 25 items in 5 domains, namely, LBP, lumbar function, walking ability, social life function and mental health,48 with the total score of 0–100, and a higher score indicates the superior condition for each domain of JOABPEQ. SF-12 is a 12-item questionnaire developed to measure the physical and mental health.49 The Cronbach’s alpha coefficients for the two subscales of SF-12 are 0.77 and 0.80, respectively.50 Notably, a consensus has been reached to apply VAS, ODI and SF-12 as the core outcome measures for clinical trials on non-specific LBP.51 The starting value, final value and change from baseline of these outcomes will be determined in each period of a cycle, and adverse reactions such as elevation of blood pressure and increased pain will be recorded as well.

Time schedule

Online supplemental table 1 presents the time schedule of participant enrolment, interventions, assessments and visits. The researcher (JZ) will recruit the participants from the Department of Orthopedics in the First Teaching Hospital of Tianjin University of TCM. During the patient screening period, JZ will inform patients interested in the trial of more details of this trial. Patients who are willing to take part in this trial should sign the informed consent form and will be assessed for eligibility in line with relevant eligibility criteria by JZ. No additional consent form will be signed because there will be no ancillary studies that involve the extraction and use of participant data and biological specimens for purposes that are separate from the main trial. Eligible patients will take a Baduanjin training course in the Department of Orthopedics guided by AFL. Patients who take Baduanjin exercise before participating in the N-of-1 trials will not be excluded in the screening stage. Afterwards, patients will undergo three treatment cycles one by one. During the first washout period of the first cycle, general characteristics such as age, gender and history of diseases will be collected by JZ. Additionally, the efficacy of previously received interventions will be eliminated if the patients receive interventions for CNLBP before participating in N-of-1 trials. Then, patients will be classified into Baduanjin exercise or waiting list group randomly at the beginning of each treatment period by WY. VAS, ODI, JOABPEQ and SF-12 are the patient-reported outcomes. Patients will be asked to answer the questions in the four scales at the beginning and end of every treatment period by TG. There are no plans for collection, laboratory evaluation and storage of biological specimens for genetic or molecular analysis in the current trial and for future use in ancillary studies. Any used drugs or adverse events will be recorded in detail. Patients who suffer from adverse reactions will be properly treated.

Data management

WeChat, one of the most widely used social networking platforms in China, has been used as the data management platform in some clinical trials.52 53 An electronic case report form (eCRF) based on WeChat will be designed to collect patient data, which can be obtained by contacting the sponsor (JZ). It is conducive to improving patient adherence and promoting data quality. During the screening period, the WeChat account of each patient will be collected. At the beginning and end of every treatment or washout period, a trained and qualified data manager (TG) will provide an unfilled eCRF to each patient via the WeChat platform. Patients will be asked to fill in the eCRF based on their own conditions. If possible, participants discontinuing or deviating from intervention protocols will be asked to fill in all electronic forms. TG is blinded to the random allocation of each patient. However, he can check the completed eCRF online in a real time manner, and contact the patients to verify and modify the questionable data if necessary. If an emergency such as adverse reaction arises and is needed to reveal the assigned intervention of a participant, TG will contact the sponsor (JZ) and the doctor (WY) to obtain the allocated intervention. When a treatment cycle is completed, individual data will be exported in time for statistical analysis. The following measures will be adopted for protecting confidentiality of potential and enrolled participants before, during and after the trial. The identification information of a participant will be replaced with an irrelevant sequence of characters. All digital files will be encrypted by TG who has access to the final trial dataset. The data used in quality control, auditing and statistical analysis will be available by filing in an application form with TG. The individual information, such as the name and mobile phone number of the patients, will be hidden during statistical analysis. Furthermore, a data and safety monitoring committee independent from the sponsor will be set up to assess the severity of the deviation from the protocol, poor compliance and severe adverse events. If necessary, patients will be discontinued from N-of-1 trials due to the above-mentioned events.

Auditing

A researcher who is employed in Tianjin University of TCM will audit the core trial processes and documents related to participant enrolment, eligibility, random allocation, patient adherence, as well as policies to protect participants by visiting the Department of Orthopedics in the First Teaching Hospital of Tianjin University of TCM, and check the data quality by browsing eCRF. The process will be performed once every month independently from the sponsor and investigators. The detected problems and suggestions will be delivered to the sponsor and investigators in writing.

Sample size

The estimation of sample size depends on multiple factors such as primary outcome and statistical analysis method. In this study, VAS, the primary outcome, will be analysed using the Bayesian hierarchical models. However, no convenient software packages are accessible for estimating the sample size of N-of-1 trials using the Bayesian hierarchical models. Therefore, we determine the sample size according to a simulation-based two-step method described by Stunnenberg et al.54 First, the simulated data of N-of-1 trials with three cycles are generated based on the following parameters. The minimal clinically important change (MCIC) refers to the smallest change of health status that leads to the clinically significant benefit in patients, such as the smallest change of VAS before and after treatment that brings about the clinically significant benefit in a particular population. Ostelo et al reported that the MCIC for chronic LBP on a VAS of 0–100 mm should be at least 20 mm.55 The minimum clinically important difference (MCID) indicates the smallest difference in health status with clinical significance between patients, like the smallest difference in VAS after treatment that is clinically significant between two groups. The present study aims to assess the difference in VAS after treatment between the Baduanjin exercise group and the waiting list group. Therefore, MCID instead of MCIC is used to estimate the sample size. A small effect on pain relief is defined as a reduction of 0.5–1.0 on a VAS of 0–10 according to the American College of Physicians guideline.4 Therefore, the MCID on a VAS of 0–10 is set to 0.5 in our study. It means that a clinically important difference is reached when the mean difference in VAS score between the two groups is more than 0.5. The SD is set to 1.0 according to our previous study.25 At the same time, the autocorrelation coefficient between two groups is set to 0.5,56 while the proportion of random missing values is set to 20%. Second, the Bayesian hierarchical model is built based on the above-mentioned simulated data by WinBUGS V.1.4.3 software. The process is repeated for 50 000 times with a burn-in of 5000 times by the Markov Chain Monte Carlo methods.57 When the simulated data from 50 N-of-1 trials are applied in building the Bayesian hierarchical models, the posterior probability of posterior mean difference >0.5 is 82.7%, which exceeds the predefined threshold of 80%. Therefore, 50 patients will be recruited.

Statistical analysis

Quantitative data will be expressed as mean with SD, while qualitative data as frequency and percentage. Data analysis will be conducted in line with the intention-to-treat principle. The missing data will be handled through the last observation carried forward. The mean differences in VAS, JOABPEQ, ODI and SF-12 score between two groups will be compared using the Bayesian hierarchical models with WinBUGS V.1.4.3 software. Additionally, the use of painkillers as a covariate will be included in the Bayesian hierarchical models to eliminate the impact of pain medications on the efficacy. Non-informative prior distribution will be used because of the lack of prior information. In addition, the number of iterations will be set to 50 000 with a burn-in of 5000 times.57 The overall and individual posterior mean differences with 95% credibility intervals between two groups will be estimated synchronously. It remains unclear whether the difference in pain relief between two groups is of clinical significance. In the sample size section, MCID is set to 0.5. Therefore, the overall and individual posterior probabilities of posterior mean difference >0.5 will be calculated. In this case, posterior probability gives a possibility that a patient achieves a clinically significant benefit. Posterior probabilities of 80% and 20% will be considered as the cut-off values to terminate the N-of-1 trial.54 When a treatment cycle is completed and individual posterior probability is more than 80%, this patient will not participate in the next treatment cycle due to the sufficient benefit, instead, he/she will be advised to perform Baduanjin exercise to improve CNLBP. If the individual posterior probability falls in between 20% and 80%, this patient will participate in the next treatment cycle because of the uncertain benefit. If the individual posterior probability is less than 20%, this patient will not participate in the next treatment cycle because of insufficient benefit and will be advised to seek alternative treatments. The sponsor (JZ) will have access to these interim results and make the final decision to terminate the trial.

Patient and public involvement

Patients and/or the public were not involved in the design, the recruitment and conduct of the study.